Re: instruction bundling (scheduling?)

andreybokhanko@gmail.com
Sun, 13 Apr 2008 10:49:18 -0700 (PDT)

          From comp.compilers

Related articles
instruction bundling (scheduling?) kevin.phillips83@yahoo.com (kphillips) (2008-04-04)
Re: instruction bundling (scheduling?) anton@mips.complang.tuwien.ac.at (2008-04-06)
Re: instruction bundling (scheduling?) kevin.phillips83@yahoo.com (kphillips) (2008-04-09)
Re: instruction bundling (scheduling?) andreybokhanko@gmail.com (2008-04-13)
Re: instruction bundling (scheduling?) kamalpr@gmail.com (IndianTechie) (2008-04-14)
Re: instruction bundling (scheduling?) kevin.phillips83@yahoo.com (kphillips) (2008-04-15)
Re: instruction bundling (scheduling?) SidTouati@inria.fr (Sid Touati) (2008-04-22)
| List of all articles for this month |

From: andreybokhanko@gmail.com
Newsgroups: comp.compilers
Date: Sun, 13 Apr 2008 10:49:18 -0700 (PDT)
Organization: Compilers Central
References: 08-04-017 08-04-025 08-04-034
Keywords: optimize, registers
Posted-Date: 13 Apr 2008 22:25:31 EDT

On Apr 9, 11:29 pm, kphillips <kevin.phillip...@yahoo.com> wrote:
> One other query - I have no specific algorithms that cater for data
> prefetches. Since a load instruction take a huge amount of cycles
> (assuming the worst case scenario - memory), is it a good idea to
> issue advanced load instructions as early as possible? Then the
> latency for loads will be assumed for quite less .. hopefully it will
> work for most cases.


...and registers' live ranges will be quite longer, leading to higher
register pressure. You have to use a heuristic here.


Regarding scheduling vs register allocation -- I agree with Anton Ertl
that it is better to do scheduling first. For two reasons:
1. Usually, optimal, unhindered (by new dependencies introduced by
register allocation) scheduling is more important (no hard data here,
just my practical observations)
2. You have quite a few registers on Itanium (especially if you
allocate on single basic blocks level and thus, have all of the
registers free again at the start of each new block)


Andrey



Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.